|User Engagement Metrics & Panda|
| 2:48 am on Apr 19, 2011 (gmt 0)|
I know there is another thread about human behavior factors which raises some interesting theories, but I'd like to focus this one specifically on quantifiable metrics, particularly avg time on site and pageviews/visit.
Please report your data in the following format:
Panda 1.0 :
Traffic Hit: -20%
Avg Pageviews/visit: +8.3%
Avg Time on Site: +13.5%
Traffic Hit: -10%
Avg Pageviews/visit: +6.8%
Avg Time on Site: +7.2%
Make sure to segment your results to Google traffic only, and that you are isolating other factors (i.e. new page designs, site improvements, etc).
We noticed these gains with Google, but didn't notice the same gains with Yahoo or Bing, which would imply that the improvements in pageviews/visit and avg time on site were isolated to the Google algo change.
There are a lot of reports of sites experiencing temporary traffic spikes before being hit by panda, which we noticed too. Perhaps Google is sending traffic to a domain, testing some user engagement metrics, and ranking the site accordingly.
Perhaps they have a way of estimating these metrics for each site based on clickbacks to the SERP pages, subsequent related searches, etc. If they can display a "Block this domain" link when a user hits the back button, then they can also easily track the time a user spends on a site before clicking the "back" button.
Has anyone else noticed similar gains in these two metrics with each update?
| 4:33 am on Apr 19, 2011 (gmt 0)|
I've been looking but haven't seen any big changes yet.
But I realized, because my site is financed by display ads, I now need to figure out how to double the average number of pageviews to get the same revenue.
| 12:15 pm on Apr 19, 2011 (gmt 0)|
Okay comparing the week before Feb. 24th to this latest week,
Our Google traffic is up 16% (other traffic not up, down in fact)
Pages Per visit from just this Google traffic has dropped 10% in this time to 3.7 per visit. Time on site is similarly affected.
This site exists in a heavily competitive and spam-filled niche, my theory is we were not penalized, or not much anyway, but the hundreds of very spammy competition was hit and we rose up this small amount kind of by default.
| 12:45 pm on Apr 19, 2011 (gmt 0)|
This whole thread is absolutely insane, as are several of the others floating around here. You're asking if Google is data mining websites in order to figure out how to rank or de-rank them? Taking just a tiny step back from that question do you see any red flags going up or can you hear any alarm bells sounding off?
Remove AdSense from your website, check one. Remove Google Analytics from your website, check two. Close your Google Webmaster Tools account, check three. Close your AdWords account, check four. Issue a personal Persona non grata to Google and if your website or web business can survive then you won't have to waste time thinking about stuff like this. If you're willing to do business with a company that f-ing sinister, then you deserve to wrestle in the dark with shadows while Google steals your lunch money. I got what I deserved doing business with Google. My eyes are now wide open, are yours?
It's time for website owners to WAKE UP.
| 4:08 pm on Apr 19, 2011 (gmt 0)|
Hi OldIrish -- Appreciate your concern, but the point of this thread isn't to decide whether or not Google data mining is a good thing, but to try to see what metrics they might be considering. It would be fairly easy for them to mine data regardless of whether or not you actually used any of their products.
Anyone else have any data they'd like to share re pageviews/avg time on site?
| 5:19 pm on Apr 19, 2011 (gmt 0)|
|Hi OldIrish -- Appreciate your concern, but the point of this thread isn't to decide whether or not Google data mining is a good thing, but to try to see what metrics they might be considering. It would be fairly easy for them to mine data regardless of whether or not you actually used any of their products. |
Anyone else have any data they'd like to share re pageviews/avg time on site?
What planet are you living on, "it would"? They are mining data, and they're likely using it in every sinister self-serving way possible. If you have a huge ecommerce site they can probably shave your organic traffic flow by an exact percentage just to get you to up your AdWords budget.
Google is pure evil and it's threads like this which attempt to parse and sort through their evil deeds that are beautiful examples of the greater psychosis that seems so predominant within many webmaster circles.
And to answer your question directly, yes all of my websites that were affected by Panda 2.0 saw major traffic spikes in March and April. Was Google gauging "user data" for quality purposes, or were they profiling user behavioral patterns in order to gauge potential conversion rates? Well seeing as though all of my ranking positions were replaced by AdSense laden scraper sites and AdSense laden big brand content farms, I'm going to place my chips on the later scenario.
| 5:34 pm on Apr 19, 2011 (gmt 0)|
Have you noticed any differences in avg pageviews/user, avg time on site immediately following those panda updates?
| 6:18 pm on Apr 19, 2011 (gmt 0)|
My network stats:
Bounce Rate: -0.21%
Average Time On Site: +23.24%
Pages Per Visit: +1.54%
What's interesting is my average time on site change. While my Google traffic losses are currently at -30% vs Pre-Panda levels, my advertising income is down by -60%. Google conveniently sniped and rerouted much of our ad clicking visitors and has actually sent them to scraper sites and low quality content farms which likely have a WAY higher CTR. If superior non-commercial sites had replaced us in our lost ranking positions then I would certainly believe that Google hit the bullseye on the quality issue, however the opposite has happened. Panda was about Google profits, with quality control being the selling point.
[edited by: OldIrish at 6:21 pm (utc) on Apr 19, 2011]
| 6:20 pm on Apr 19, 2011 (gmt 0)|
Absolutely they are. They have been collecting data for years. You can't run any type of numerical model (especially statistical algorithms that define 'machine learning') without data --- and LOTS of it. The more you get, the bigger your statistical dataset, and the more the machine can learn about expected results. I'm not convinced that GA has anything to do with it though, since my sites have never used GA. Google has so much user data at its disposal that it would be mind boggling.
|Perhaps they have a way of estimating these metrics for each site based on clickbacks to the SERP pages, subsequent related searches, etc. If they can display a "Block this domain" link when a user hits the back button, then they can also easily track the time a user spends on a site before clicking the "back" button. |
One of the simplest metrics they could calculate. They have to know about the user's return to the SERPS in order to serve the "Block" option. They know how long it took the user to return, and they can calculate a probability distribution (used in statistical modeling) from random variable testing. This could explain traffic surges and temporary rankings we are seeing for some terms.
| 6:28 pm on Apr 19, 2011 (gmt 0)|
Thanks for sharing. Let's try to stick to these two metrics for the purpose of this thread (I think we all agree they are mining data -- but let's leave that for another discussion).
So far it looks like:
Panda Losers (2): Increased pages/visit, increased avg time on site
Panda Winners (1): Decreased pages/visit, decreased avg time on site
Only 3 data samples so far, though. Anyone else have data on these two metrics to share?
| 6:31 pm on Apr 19, 2011 (gmt 0)|
|Let's try to stick to these two metrics for the purpose of this thread |
We're just trying to offer information/experience that hopefully contributes to the discussion (in addition to some stats), to help address the OP. We're only trying to help. :)
::Edited to be friendlier. The bad mood hit me at lunch :) I also provided some basic stats in my post below.
[edited by: crobb305 at 7:16 pm (utc) on Apr 19, 2011]
| 6:36 pm on Apr 19, 2011 (gmt 0)|
Actually, I am the OP. I registered a long time ago with two usernames and couldn't remember the original logins. If a moderator could change all my Ackkster posts to "Ackk", and delete my "Ackkster" account that would be great. Sorry for the confusion.
And the reason I want to focus on those two metrics, and only those two metrics, is because I've seen too many other threads get hijacked into general Panda discussion. I'd like to focus on only these two metrics for the purpose of this thread.
| 6:38 pm on Apr 19, 2011 (gmt 0)|
I knew you were the OP. It was obvious from the names. I was just specifically addressing the suppositions in your OP, and I was in the process of calculating some stats for you. You have to be patient. It takes time to type and compute. :)
| 6:42 pm on Apr 19, 2011 (gmt 0)|
6.8 pageviews per visit on a video site is low while on a guide site it's high. Google not only measures your site they do so against similar sites in order to create rank. For that reason your metrics are only valid against similar sites so comparing them to "everyone else" is of limited value.
| 6:45 pm on Apr 19, 2011 (gmt 0)|
crobb - I also didn't see the reply concerning SERP clickbacks in your first post when I first read it, sorry about that! Just trying to make sure this thread stays somewhat focused :)
| 6:49 pm on Apr 19, 2011 (gmt 0)|
sgt_kickaxe - good point. I guess I am not interested in the exact numbers, per se, but in whether or not there was a % gain or loss, for the same site, after Panda 1.0/2.0. Not trying to compare across sites or industries.
| 6:53 pm on Apr 19, 2011 (gmt 0)|
Panda 1.0 :
Traffic Hit: -50%
Avg Pageviews/visit: +10%
Avg Time on Site: +12%
Traffic Hit: -8%
Avg Pageviews/visit: +9%
Avg Time on Site: +5%
Wordy Disclaimer: These data are based on overall traffic trends. I don't see a noticeable improvement on Google-only traffic (mainly because I get so little after Panda 1). When Google added features last year (auto-complete, etc), my bounce rate soared. I blame those features because for 6 years my bounce rate was under 32% (85% of my traffic comes through the homepage). After the features were implemented, my bounce rate soared to 38% to 40%. I believe my bounce-rate improvements post Panda 1 and Panda 2 are because of better traffic from Bing/Yahoo.
After Panda 2, I saw my bounce rate drop to its lowest point in a year (26%) for a few days. Now that Google has lost my traffic, I don't know how much data they can mine from users who search on Bing/Yahoo (unless they have Toolbar installed or stay logged into their Google account). I do not credit Google with improvements in my bounce rate; I credit Bing/Yahoo for providing better results. I do credit Google for killing my bounce rates in 2010.
| 10:02 pm on Apr 19, 2011 (gmt 0)|
|I guess I am not interested in the exact numbers, per se, but in whether or not there was a % gain or loss, for the same site, after Panda 1.0/2.0. |
I don't see the point - there's not even any proof from the data I'm looking at at that it is even site-wide. I have data that clearly shows its an individual/set of page(s) problem
I haven't seen any comment on whether there even was a Panda 2 outside the USA, just a "BIG panda"( trademark claimed !)
i.e. the big picture will not explain , or expose, the multi-variable nature of this update.
All sites are different , user generated content, ecommerce, info sites, affiliate sites, MFA, discount voucher codes, news - they will have have been hit/not hit in different ways.
I even had someone on the phone today talking about Fmarketing ( Facebook ) !
| 10:46 pm on Apr 19, 2011 (gmt 0)|
For the period 2/2 through 2/23:
92,862 Google visits
1:52 time on site
For the period 3/28 through 4/18:
37,062 Google visits
1:55 time on site.
Not much difference (except for the number of Google visitors, of course).
| 12:10 am on Apr 20, 2011 (gmt 0)|
|I don't see the point - there's not even any proof from the data I'm looking at at that it is even site-wide. I have data that clearly shows its an individual/set of page(s) problem |
I think the OP wanted to kind of judge whether an improvement in bounce rates could be tied to other sources (e.g., Bing/Yahoo) after the loss of Google traffic, or if people are seeing a specific improvement in post-Panda traffic still coming from Google. For my site, when I lost about 60% from Panda, my bounce rate improved dramatically. I chalk it up to a better quality of traffic coming from Bing/Yahoo. Others might be able to nail it down to better longtail traffic coming from Google (i.e., loss of trophy phrases).
Also, regarding the sitewide effect, I think it's most noticeable for people (like me) who tend to get the majority of their traffic through the homepage. Historically, about 80% of my traffic comes through the homepage. My site isn't built with categories or subdirectories, so I don't see a lot of internal linkage. Most people tend to just link to my homepage. I did see a 60% drop in traffic that normally came through the homepage, and homepage rankings dropped 100 to 500 positions for its main phrases. I've been able to nail the culprits down to about 5 hard-hit internal pages, so it does seem to be some sort of phrase-based scoring and backwards bleeding of penalty to the homepage in my situation. I'm sure there's a spectrum of possible penalty paths through the site depending on the linking structure, so I can't speak for everyone.